[lmi] Patching severl xml-library defects

[Greg Chicares]

Evgeniy--Could I ask you to

 - prepare patches as described below,
 - place them on our site
     (I guess our "patch tracker" is good enough for now), and
 - eventually submit them to the maintainers?

I've written (1) through (3) by hand. You might find a better way
to fix those problems. Of course, the maintainers might choose
not to fix the problems at all:

  "...then don't use msw."
  "...but every program should link to glibmm."
  "...just ignore those warnings."

but it's worth a try.

What (0) fixes is a plain error. The others are errors with
compiler options like
  -pedantic-errors -Werror
in which I delight.

(0) libxslt-1.1.17/libexslt/Makefile.in
    libxslt-1.1.17/libxslt/Makefile.in

[the patch you emailed me this weekend to make shared-library
builds work on msw]

(1) libxslt-1.1.17/libxslt/xsltInternals.h

-     XSLT_FUNC_EXTENSION,
+     XSLT_FUNC_EXTENSION
  #ifdef XSLT_REFACTORED
+                        ,
      XSLT_FUNC_OTHERWISE,
      XSLT_FUNC_FALLBACK,
      XSLT_FUNC_MESSAGE,
      XSLT_FUNC_INCLUDE,
      XSLT_FUNC_ATTRSET,
      XSLT_FUNC_LITERAL_RESULT_ELEMENT,
-     XSLT_FUNC_UNKOWN_FORWARDS_COMPAT,
+     XSLT_FUNC_UNKOWN_FORWARDS_COMPAT
  #endif

This prevents:
| In file included from C:/usr/local/include/libxslt/documents.h:17,
|                  from C:/lmi/src/lmi/ledger_text_formats.hpp:33,
|                  from C:/lmi/src/lmi/illustration_view.hpp:40:
| C:/usr/local/include/libxslt/xsltInternals.h:401: error: \
| comma at end of enumerator list

(2) libxml++/parsers/parser.h

  #ifdef WIN32 //TODO: Why do we do this? murrayc.
- #pragma warning (disable : 4786)
+ //#pragma warning (disable : 4786)
  #endif

Obviously the maintainers would be more likely to accept a
patch that adds appropriate conditionals:
  #ifdef MSVC_SOMETHING
  ...
  #endif // MSVC_SOMETHING

I didn't keep a copy of the gcc error message, but the problem
is of course an unrecognized (and nonstandard) pragma.

(3) libxml++/validators/validator.h

Same problem as (2) above.